Analysis of diffuse parenchymal liver disease by liver scintigrams: differential diagnosis using neuro and fuzzy

In colloid liver scintigraphy, diagnosis of diffuse parenchymal liver disease such as chronic hepatitis or cirrhosis is evaluated by size and distortion of the liver, distribution of tracer in the liver, size and activities of tracer in the spleen, visualization of the bone marrow and so on. It is not difficult to read a scintigram which shows a typical pattern of normal, chronic hepatitis and cirrhosis; however in some cases it is difficult to distinguish normal or chronic hepatitis and chronic hepatitis or cirrhosis visually. Therefore, we tried to use fuzzy inferences to perform differential diagnosis in chronic hepatitis (CH), severe fibrosis (SF) and liver cirrhosis (LC). First, five features in colloid liver scintigrams were measured or evaluated visually. These features were liver size index (left lobe/right lobe), splenomegaly, the degree of visualization of the bone marrow, liver deformity, and distribution of tracer in the liver. Having fuzziness in these data, certain characteristics of these features were considered to be fuzzy sets and thus could be expressed in membership functions. Fuzzy inference was carried out using these data and fuzzy rules. Using fuzzy inference, differential diagnosis in LC could be performed up to 100%, but those of CH and SF could not be performed sufficiently. Using neural network CH, SF and LC could be diagnosed up to 63%, 80%, and 88%, respectively. But fuzzy inference had the merit to evaluate the degree of disturbance by the center of gravity of the resulting membership function. Therefore by combining neural network and fuzzy inference, CH, SF, and LC could be differentiated to the degree 77%, 80%, and 100%, respectively.